Which multi-agent framework and LLM did you use exactly?

NL2Code / CodeR

152 stars 17 forks source link

Which multi-agent framework and LLM did you use exactly? #6

Closed Slach closed 5 months ago

Slach commented 5 months ago

Is this not clear from paper Are you going to publish it?

NL2Code commented 5 months ago

Hi, @Slach, we implement CodeR with multi-agent based on the SWE-agent code, and do not use other framework. The LLM used in CodeR is GPT 4 preview-1106 as we have already stated in Section 4.1.1 of our paper. For your second question, we will continue to refine CodeR and decide whether to publish as appropriate.